You mean you outcome result/dependent variable data is on a likert scale? Well you can use it as is, or you can dichotomise the variable. You can dichotomise based on a particular logic that you have in mind E.g. point 1-3 and then 4-5, or you can look at the frequencies of the variable on each of the points and insert a cut off point halfway where each of the two categories would have more or less similar points. So develop many version of your outcome and test that with your independent variables - play around with the data and see which one can best interpret your hypothesis.